JP2000172654A - Distributed object-managing system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable performance improvement based on the load distribution of an entire system and to attain improvement in the utilization efficiency of the entire system, even when increasing/decreasing the users of services provided by plural programs in the system in a distributed object environment. SOLUTION: For a server computer 1, for example, the copy of a server object is prepared dynamically during the operation of a server program 12A, transferred onto another server computer 2, for example, of low CPU utilization rate and operated and corresponding to a remote method call from a client computer, remote method calling is made to perform by the copied object. When the number of times of remote method calling to the copied object does not exceed a prescribed number of times within the prescribed time, the utilization efficiency is regarded as being low, and the copied object is automatically eliminated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention relates to a part of a program constituting a program when programs running on computers on various networks such as the Internet and an intranet communicate using distributed object technology. When creating a copy of an object dynamically according to the administrator's settings and transferring it to another computer for operation, the number of connections within a predetermined period, the upper limit of the CPU utilization of the computer on which the copied object runs, etc. The present invention relates to a distributed object management system having a function of automatically deleting a copied object in accordance with a policy of a system administrator, thereby improving the use efficiency of the entire system.

[0002]

2. Description of the Related Art Conventionally, a server duplication technique has been used in order to distribute the load to a plurality of computers constituting a distributed system and to improve the utilization efficiency of the computers in the entire system. According to this technique, the entire program that provides the service runs on the computer on which the program runs, and all services can be used simultaneously on a plurality of computers.

In a distributed object environment,
Japanese Patent Laid-Open No. 10-207849 discloses a method of preparing a duplicate object in advance and making it possible to use the duplicate object in the event of a failure to perform high reliability and load distribution. Methods ".

[0004]

However, in the prior art using the server replication technology, since the entire program for providing the service is running from the beginning, if the number of users is small, useless services are not provided. There is a problem that an operating state occurs, and as a result, utilization efficiency of the entire system is reduced.

On the other hand, in the prior art disclosed in Japanese Patent Application Laid-Open No. 10-207849, it is necessary to create a duplicate object in advance, so that the system uses unnecessary resources similarly to the above-described server replication technology. And, as a result, there is a problem that the utilization efficiency of the entire system is reduced.

SUMMARY OF THE INVENTION An object of the present invention is to provide a system in a distributed object environment in which a plurality of programs composed of objects communicate using a distributed object technology. It is an object of the present invention to provide a distributed object management system which can improve the performance by distributing the load of the entire system and improve the efficiency of using the resources of the entire system even if the number is small.

[0007]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention provides a server object, which is a distributed object that is distributed on a network and constitutes a server program, and is transferred to another computer. In a distributed object environment system comprising: a plurality of server computers each including a processing unit; and a management computer including a naming service processing unit that manages position information of a server object running on the plurality of server computers. Each computer stores connection limit management information including any or all of the upper limit of the number of simultaneous method calls to the server object in the own computer, the upper limit of the CPU usage rate, and the upper limit of network traffic per hour. Management information for connection restriction File is provided, and the management computer is inquired at the time of an acquisition request for reference to the server object from the client computer via the network whether the requested server object can be connected to the operating server computer, When a connection-disabled response is received from the server computer due to exceeding the upper limit value specified in the connection restriction management information, a request to create a copy of the server object requested by the client computer is transmitted to the server computer. Naming service processing means for transferring a reference to a duplicate object created by the duplication request to another server computer;
Call monitoring means for acquiring the connection restriction management information of the server computers on which the copy source and copy destination server objects are running, and managing calls to the copy source server object and the copy objects of each server computer. It is characterized by having.

[0008] In brief, according to the present invention, a plurality of server computers, a server program composed of server objects distributed on the server computer, a management computer for managing the server program, and each server computer are provided. Assume that you have a network to connect to.

On each server computer, a duplicate object for duplicating a distributed object which is a part of a server object constituting a server program, transferring the duplicate object to another computer, and deleting an active duplicate object. The management program is running. Also, a policy management program that manages a server program load upper limit management policy set by a system administrator, such as an upper limit on the number of simultaneous method calls to each distributed object and an upper limit on the CPU utilization rate of the computer itself, is running.

On the management computer, a distributed object replication management program for instructing the replication of the server object, a policy monitoring program for selecting a connection destination object from information of the policy management program on each server computer, and a server object (replication) Original
Naming service program for managing location information of (copy destination), server computer for providing the service, client computer for using the service, network management program for managing the configuration of the network to which they are connected, and server for providing the service A system monitoring program for collecting the CPU utilization of the computer is running.

When a server program running on each server computer is started, a reference to the server object (information such as an ID for specifying the server object) is registered in a naming service program on the management computer. The policy management program obtains a connection restriction policy for the activated server object.

When a client computer attempts to invoke a method on a server object using the distributed object technology, the client computer first requests the naming service program on the management computer to obtain a reference to the server object. Upon receiving the reference request, the naming service program inquires of the policy monitoring program whether or not connection is possible. If the connection limit has been reached, the naming service program issues a replication creation request to the distributed object replication management program.

The distributed object copy management program that has received the copy creation request selects a server computer with an adjacent (or short distance) network path from the client computer based on the network configuration obtained from the network management program, A request for duplication of the object is issued to the duplicate object management program on the server computer on which the server object is running, a reference to the duplicated server object is obtained, and the reference is returned to the naming service program.

The naming service program registers the reference of the duplicate object in the reference management database managed by itself, and returns the reference to the client program.

[0015] The client program performs a method call for the received reference, thereby performing the communication between the distributed objects as usual and obtaining the result.

Further, the duplicate object management program on each server computer checks the log stored in the external storage device every time the object is called,
Duplicate objects that have not been connected a predetermined number of times or more within a certain period are deleted from the memory.

By employing the above method, when a plurality of calls are made to a server object which is a distributed object constituting a service program, the server object is automatically copied in accordance with the load state of the server computer. Is created on another computer, and the utilization efficiency of the entire distributed system can be improved.

Further, since a server computer adjacent to or close to the client computer is selected as a copy destination according to the network configuration, network traffic can be reduced. In addition, since the copy object having a low usage rate is automatically deleted from the memory, it is possible to prevent the resource of the server computer from being wasted.

[0019]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described.
This will be described in detail with reference to the drawings. FIG. 1 is a system configuration diagram showing an embodiment of a distributed object management system to which the present invention is applied. The system of this embodiment includes a plurality of server computers 1 and 2 and a management computer 3, which are connected by a network 4.

The server computer 1 has a CPU 11A,
A terminal device 11 comprising a memory 11B, an external storage device 12 storing a server program 12A, a duplicate object management program 12B, a policy management program 12C, a connection restriction policy description file 12D, a call log file 12E, and a communication port 13. It is configured.

Similarly, the server computer 2 has a CPU
21A, a terminal device 21 comprising a memory 21B, a server program 22A, a duplicate object management program 2
2B, an external storage device 22 storing a policy management program 22C, a connection restriction policy description file 22D, and a call log file 22E, and a communication port 23.

The management computer 3 includes a terminal device 31 including a CPU 31A and a memory 31B, a naming service program 32A, a distributed object copy management program 32B, a policy monitoring program 32C, a network management program 32D, and a system monitoring program 3.
2E, an external storage device 32 in which a reference management database 32F and a network configuration file 32G are stored;
And a communication port 33. Here, the server programs 12A, 22 of the respective server computers 1, 2
A client computer (not shown) that receives the service provided by A is connected to the network 4 directly or via another network.

FIG. 2 shows a reference to a distributed object (an ID for specifying a distributed object) used by the naming service program 32A running on the management computer 3.
FIG. 3 is a diagram illustrating an example of a table configuration of a reference management database 32F storing information such as the name of a distributed object 201 and the name of the original server computer on which the distributed object is running (or IP address) 202, the name (or IP address) of the server computer on which the copy is currently running
Column 203 and a column 204 of the name (or IP address) of the server computer whose copy has been deleted.

The server names 203 and 204 are such that the names or IP addresses of a plurality of server computers are “blank”
Or "," (comma).

FIG. 3 is created by the administrator of the distributed system,
It shows an example of a connection restriction policy description file (12D, 22D) used by the policy management programs 12C, 22C running on each of the server computers 1, 2, and the first line shows one line per minute (per predetermined time). The upper limit number of connections 301, the upper limit 302 of the CPU usage rate on the second line, and the upper limit 303 of network traffic per minute (per predetermined time) are described on the third line.

FIG. 4A shows server computers 1 and 2, client computers, collected by the network management program 32D running on the management computer 3.
FIG. 14 is a diagram illustrating an example of a network configuration file 32G expressing a network connection relationship. In the network configuration file 32G, the name of a router (eg, Router1-1) that is a route is defined at the beginning of the first line, and thereafter, devices (routers, hubs, computers, etc.) that can be reached from the router in one hop. Name (eg Rou
ter1-2, WS1, WS2, etc.) or IP addresses are defined side by side, separated by spaces. From the second line onward, a router serving as a route (for example, Router1-
A router that can be reached in 1 hop from 1) (eg, Rout
er1-2) is defined at the top, and the names of the devices (for example, W) excluding the router described at the beginning of the line among the devices that can be reached by one hop from the router thereafter
S3, PC4, etc.) or IP addresses are defined side by side, separated by spaces. Similarly, the subsequent lines are lined with the first line of routers that can be reached by two hops from the root router, and are repeatedly defined until there are no more routers that can be reached by one hop.

Usually, a network can loop, but a router is defined to appear only once at the beginning of a row.

FIG. 4B illustrates the connection relationship defined in the network configuration file 32G of FIG. 4A. In FIG. 4B, WS1 to WS6 are workstations, and PC1 to PC10 are personal computers.

FIG. 5 shows a call log file (12E, 22E) storing a log of remote method invocation for each distributed object used by the duplicate object management programs 12B, 22B on each of the server computers 1, 2.
FIG. 17E is a diagram illustrating an example of E), and a log at the time of one connection is described in one line. That is, one line describes the date and time 501 when the method of the duplicate object is called, the name (or ID) 502 of the called duplicate object, and the name (or IP address) 503 of the calling client computer. I have.

The operation of this embodiment will be described below with reference to a flowchart. FIG. 6 is a flowchart showing the operation of the naming service program 32A running on the management computer 3.

In FIG. 6, when the naming service program 32A is started, it waits for a connection from a server application or a client application (step 601). Next, when there is a connection from the server application or the client application, it is determined whether or not the connection is a server application (step 602). In the case of a server application, a reference to a server object is obtained from the server application, and the server name is obtained. Figure 2 and reference of server object
Is stored in the reference management database (DB) 32F (step 603). When the storage is completed, the server application is notified of this (step 604), and thereafter the connection is disconnected (step 605), and the processing is repeated from step 601.

On the other hand, in the case of connection from the client application, it is confirmed whether or not the reference of the requested server object exists in the reference management database 32F (step 606). If not, an error is returned (step 606). 607), and repeat from step 601. However, if there is a reference, the reference management database 32 determines whether there is a server computer on which a duplicate object of the requested server object is running.
F is determined by the server 203 that is operating the replication (step 6).
08) If there is no server computer on which the duplicate object is running, the policy monitoring program 32C
The server inquires of the server object whether connection is possible (step 609), and determines whether or not connection is possible based on the response (step 610). If connection is possible, the server object is referred to. Return to the client application (step 611) and repeat from step 601. However, if the connection is not possible, the distributed object replication management program 32B
Request the duplicate creation of the object by passing the name of the server object to be duplicated, the name of the server computer on which the server object is running, the name of the client computer trying to connect to the server object, and the name of the duplicated server computer. (Step 612), a reference to the created duplicate object is obtained (Step 613), and the name or IP address of the server computer where the duplicate is created is stored in the reference management database 32F of FIG. 2 (Step 614). Is returned to the requesting client computer (step 615), and the process is repeated from step 601.

On the other hand, if there is an active copy object, the policy monitoring program 32C determines whether or not it is possible to connect from the server computer closest to the requesting client computer based on the network configuration file 32G obtained from the network management program 32D.
(Step 616), it is determined whether or not connection is possible based on the response result (step 617). If there is a duplicate object that can be connected, a reference to the object is returned to the requesting client computer (step 618), and step 601 is performed. Repeat from If there is no connectable duplicate object, the process is executed from step 612.

FIG. 7 is a flowchart showing the operation of the distributed object copy management program 32B running on the management computer 3.

In FIG. 7, when the distributed object replication management program 32B is activated, the network configuration program 32D shown in FIG.
G is obtained (step 701). Next, it waits for a connection from the naming service program 32A (step 7).
02). When connected from the naming service program 32A, the name of the server object to be copied, the name of the server computer on which the server object is running, the name of the client computer trying to connect to the server object, and the name of the server computer that has been copied The name is obtained (step 703). Next, from the network configuration file 32G, a server computer which is located close to the client computer requesting the server object and has not been copied is selected (step 704), and the copied object of the server computer on which the server object is running is selected. It connects to the management program (12B or 22B) and sends the name of the server object to be copied and the name of the server computer to which the copy is to be transferred (step 705). When the copying is completed, a reference to the copied server object is obtained (step 706), and this is returned to the naming service program 32A (step 707), and the process is repeated from step 702.

FIG. 8 is a flowchart showing the operation of the policy monitoring program 32C running on the management computer 3. In FIG. 8, when the policy monitoring program 32C starts, the naming service program 3
Wait for a connection from 2A (step 801). When the connection is made from the naming service program 32A, a server computer name and a server object name are received (step 802). Next, the received policy management program (12C or 2C) running on the server computer
2C) (step 803), transmits the server object name (step 804), receives a result indicating whether connection to the server object is possible (step 805), and returns the result to the naming service program 32A (step 803). 806), and repeat from step 801.

FIG. 9 is a flowchart showing the operation of the network management program 32D running on the management computer 3. In FIG. 9, when the network management program 32D is started, SNMP (Simple Network
A server computer or a client computer connected to the network 4 is called up using the P.K. Management Protocol) or PING to create the network configuration file 32G shown in FIG. 4A (step 90).
1). Thereafter, the process waits for a connection from the naming service program 32A or the system monitoring program 32E (step 902), and if connected, transfers the network configuration file 32G (step 903) and repeats from step 901.

FIG. 10 is a flowchart showing the operation of the system monitoring program 32E running on the management computer 3.

In FIG. 10, when the system monitoring program 32E starts, the network management program 32E
D is obtained from D (step 1001). Then, CPU utilization is collected for the server computers 1 and 2 on the network configuration using SNMP (step 1002). Next, it is determined from the policy monitoring program 32C whether there is a connection (step 1003). If connected, the CPU utilization of the requested server computer is returned (step 1004), and the process is repeated from step 1002. If there is no connection, the process is repeated from step 1002 after a predetermined time has elapsed.

FIG. 11 shows a duplicate object management program 12B, 2 running on each of the server computers 1, 2.
It is a flowchart showing the operation of 2B. In FIG. 11, a duplicate object management program (12B, 22
When B) starts, the distributed object copy management program 32B on the management computer 3 or the copy object management program (22B,
Wait for a connection from step 12B) (step 1101). If the connection is made from any of the programs, it is determined whether the connection is from the distributed object copy management program 32B or a connection from the copy object management programs (22B, 12B) on another server computer (step 1102). If the connection is from the distributed object duplication management program 32B, the name of the server object to be duplicated and the name of the server computer to which the duplicated server object is to be transferred are obtained (step 1103). Next, a connection is made to the copy object management program (22B, 12B) on the transfer destination server computer (step 1104). Thereafter, the active server object is converted into, for example, a binary code conforming to the data format of the communication path (step 1105), and the program data of the server object stored in the external storage device of the own computer and the date and time of creation of the program data are stored. They are transferred together (step 1106). Thereafter, a reference to the transfer destination server object is obtained (step 11).
07), the reference is returned to the distributed object replication management program 32B (Step 1108), and Step 11
Repeat from 01.

In the case of connection from a duplicate object management program (22B, 12B) on another server computer, the binary code of the server object, the program data of the server object, and the creation date and time of the program data are obtained (step 1109). The program data is stored in the external storage device of the own computer with the date of creation date and time (step 1110), and a server object is generated from the received binary data (step 111).
1), the reference of the server object is returned (step 1112), and the process is repeated from step 1101.

FIG. 12 is a flowchart showing the operation of the policy management programs 12C and 22C running on each of the server computers 1 and 2. In FIG. 12, when the policy management programs (12C, 22C) are started, the connection restriction policy description files (12D, 22D) are read (step 1201). Thereafter, it is checked whether there is a connection from the policy monitoring program 32C on the management computer 3 (step 1202). If there is a connection, a server object name for checking whether connection is possible is received (step 1203). The call log file (12E, 22) stored in the storage device
From E), a connection log for the corresponding server object is read for a certain period (step 120).
4) If the connection limit of the policy has not been reached, the connection is returned to the policy monitoring program 32C if the connection is possible;
Repeat from 02.

If there is no connection from the policy monitoring program 32C, it is checked every predetermined time whether or not the connection restriction policy description file (12D, 22D) has been changed (step 1206). (12D, 22D) are read (step 1207), and the process is repeated from step 1202.

The present invention is not limited to the above-described embodiment, but can be implemented by variously changing each part as needed. For example, when a duplicated object is transferred to another server computer, a snapshot of the object is stored in an external storage device on the duplicated server computer, and the snapshot is deleted even if the duplicated object is deleted. The next time the replica is created, the version of the source server object and the snapshot are checked. If the versions are different, normal duplication processing is performed. If the versions are the same, the server object is deleted. It is possible to create a duplicate object from a snapshot stored in the external storage device without performing the duplication process. This reduces the traffic for creating a duplicate object, which is more effective.

[0045]

As described above, when the present invention is used, a copy of a server object is automatically generated, and a server object on a server computer with a small load can be used without making any change on the client program side. And the execution efficiency of the distributed system can be improved. Also, by monitoring the connection ratio and automatically deleting the duplicated server object, it is possible to prevent the duplicate object from wasting computer resources.

[Brief description of the drawings]

FIG. 1 is a system configuration diagram showing an embodiment of a distributed object replication management system according to the present invention.

FIG. 2 is an example of a table configuration of a database storing references to distributed objects used by a naming service program running on a management computer in the embodiment of FIG. 1;

FIG. 3 is a diagram showing an example of a connection restriction policy description file used by a policy management program running on each server computer in the embodiment of FIG. 1;

FIG. 4 is a diagram showing an example of a network configuration file used by a network management program running on a management computer in the embodiment of FIG.

FIG. 5 is a diagram showing an example of a call log file storing a log of a remote method call for a distributed object on each server computer in the embodiment of FIG. 1;

FIG. 6 is a flowchart showing an operation of a naming service program running on a management computer according to the present invention.

FIG. 7 is a flowchart showing an operation of a distributed object copy management program running on a management computer in the embodiment of FIG. 1;

8 is a flowchart showing the operation of a policy monitoring program running on a management computer in the embodiment of FIG.

9 is a flowchart showing an operation of a network management program running on a management computer in the embodiment of FIG.

10 is a flowchart showing the operation of a system monitoring program running on a management computer in the embodiment of FIG.

11 is a flowchart showing the operation of a duplicate object management program running on each server computer in the embodiment of FIG.

FIG. 12 is a flowchart showing the operation of a policy management program running on each server computer in the embodiment of FIG.

[Explanation of symbols]

1, 2 ... server computer, 3 ... management computer,
4 Network, 11, 21, 31 Terminal device, 11
A, 21A, 31A CPU, 11B, 21B, 31B
... Memory, 12, 22, 32 ... External storage device, 13, 2
3, 33: communication port, 12A, 22A: server program, 12B, 22B: duplicate object management program, 12C, 22C: policy management program, 12
D, 22D: Connection restriction policy description file, 12E,
22E: Call log file, 32A: Naming service program, 32B: Distributed object replication management program, 32C: Policy monitoring program, 32D ...
Network management program, 32E: System management program, 32F: Reference management database, 32G: Network configuration file.

Claims (3)

[Claims] 1. A plurality of server computers which are distributed and arranged on a network and have a processing means for duplicating a distributed object constituting a server program and a processing means for transferring the distributed object to another computer, and running on a plurality of server computers. In a system of a distributed object environment comprising: a management computer having a naming service processing means for managing location information of a distributed object; wherein each of the server computers executes a simultaneous method invocation on a server object which is a distributed object in its own computer. Number limit, CPU usage limit,
A connection restriction management information storage file for storing connection restriction management information including any or all of the upper limit value of network traffic per hour, wherein the management computer refers to a server object from a client computer via a network At the time of the acquisition request for the server object, inquires whether the requested server object can be connected to the operating server computer, and if the server computer exceeds the upper limit specified in the connection restriction management information, the connection is disabled. When a response is received, a request to create a copy of the server object requested by the client computer is transmitted to the server computer, and a reference to the duplicate object created by the copy request is transferred to another server computer. And the connection limiting information of the server computer on which the replication source and the replication destination server object are running, and manages the call of each server computer to the replication source and the replication destination server object. A distributed object management system comprising a call monitoring means. 2. Each of the server computers,
2. The distributed object replication management system according to claim 1, further comprising means for automatically deleting the replicated server object when the number of calls to the replicated server object does not exceed a predetermined number in a predetermined period. 3. When the duplicated server object is transferred to another server computer, a snapshot of the duplicated object is stored in an external storage device on the duplicated server computer, and the duplicated server object is deleted. In the event of a snapshot, the snapshot is retained without erasing, and the next time the replica is created, the version of the server object and the snapshot in the source server computer are checked. 3. The distributed object management system according to claim 2, wherein the processing is performed, and if the versions are the same, the duplicate object is created from the snapshot stored in the external storage device without duplicating the object. .